Ask HN: Complete Deepfake Nudes – The Future?

16 points by blunte ↗ HN
This might be a bit controversial for HN, but it's safe to assume that some HN readers have been on the internet before.

For those who are familiar with some of the Reddit groups, you may have noticed a shift from simple sharing of photos to promotional sharing which leads to an OnlyF* type site where there is an economic goal. Nothing wrong with this, and not judging; just an observation.

So the question is, if we can generate incredibly detailed and believable faces, why not entire bodies with faces? This must exist already.

Who is working on a system to build adult content, made to order, that doesn't involve actual humans? We can already produce very believable landscapes, structures, and moving objects. How long until the carnal content is algorithmically produced?

18 comments

[ 3.1 ms ] story [ 51.8 ms ] thread
> How long until the carnal content is algorithmically produced?

I'll go you one further,

how long until sexual function is algorithmically interpreted and expressed? We may reach the point when human sexual function is seen, or even institutionalized, as a useful zone of concern for e.g. meaningful productivity output toward larger goals.

It used to be that you'd watch movies (or work in a place, IRL) where the businessman would work all day and then head to the club. Instant exhaustion relief.

But today's work environments are merging with the club, which IMO has been a pretty nice idea and a good way to build on psychological feedback loops. Early in COVID there were lots of funny WFH videos where sensory home life merged with office life, as well.

And what is sexual interest--what is sexual interest and activity doing for us? Can we learn to constructively stimulate sexual levels of interest toward some species-level goal beyond simple reproduction?

_Algorithmically so_?

(IMO a huge portion of society will not allow itself to consciously access this zone of thought, let alone accept it. Not just for religious reasons, but for overriding preferences reasons. There are big, important things that are more important to them than sex. But sex is still vital to humans, and we can't let one group's preference push helpful things away just because they're not used to dealing with them)

(2. I could see something like a sexual captcha in a mildly-dystopian future. OK you want free access to a few hours of our service, so select those that most excite you, anonymous person? And instead of stoplights or boats or bicycles in this captcha, you get some REALLY amazing images. Some are wholly, overtly sexual. Others--uncomfortable but FASCINATING?! What...? Why...? Nobody really knows; the algorithm is just algorithming, but here is an energy source, is nobody really expected to tap into it?)

Anyway thoughtful post, appreciated

I'm intrigued by your reply, but admittedly I don't understand some of what you're saying...

As for the captchas, I have noticed some obviously (poorly) AI-generated images... vehicles with really warped wheels (like the ears of some AI faces), and other tell-tale signs of unreality.

But expanding on this general idea, we all have built in concepts of beauty and sex and coupling. Can this be distilled to some essence which could be represented in very non-human ways but still produce the same emotional response?

A horny man (and from what I read, a horny woman also) can project sexual imagery over even the most mundane objects and movements. A piston in an engine? A pressure valve in a hydraulic system? I'm being a bit absurd, but humans are quite good at pattern matching in strange situations.

If someone can figure out how to provide sexual stimulation without being considered pornographic, this could be ... I cannot think of an expression which would not be gratuitously sexual. It would be a big deal. Nope, that was a bit sexual. But you get the idea.

Yeah. It's really subjective and intuitive, it requires suspension of disbelief, maybe pants also, but the cultural progression is so clear,

I mean just when you get done casting a rather masculine form of effeminate space hero to wide acclaim--OK you threw in a kitty cat, but still--you start to get feedback from keyed-up audience members to show more of the android, too. Or IDK, those aliens themselves looking kind of interesting. And spaceships, we are starting to worry ourselves now because are these metaphors really us, or what is reality, it's all kind of uncomfortable because I know what is real...and here is where interesting thought models go to die, this R-word. The word that makes us take ourselves so seriously that the S-word is thought of as a mockery of our R-word, because in R-life, the S-word is only transacted in weaker moments or whatever.

The most interesting model for this to me is simply sensory variety and juxtaposition with developmental implications. E.g. once you find this place of security, you want more, but _different_ more. Maybe you're not ready to admit it, and what it looks like, or COULD look like, but I think an algorithm could easily blow past questions like this because who are you going to blame for these machinistic extrapolations which also beckon waves of deep pleasure. Something like that could totally leapfrog those questions, similar to the way Kurgesagt recently and cleanly leapfrogged my valued position as the family science teacher. Was it Kurgesagt, or was it Youtube? The internet? Who causes what? Well, here we are, anyway.

I think maybe even the topic of sexuality could be leapfrogged, go away with a whimper but ideally it also brings the whimper forward and builds upon it, despite any groans from those disinclined few...and then gone, and here we are and why use that old word which is now a valued form of intrinsic energy available to humankind...another old word...

Well, I feel somewhat infantile and not understanding many of these words, but still grasping the concept.

I did get this: you want more, but _different_ more

This is absolutely true. We thrive on new experiences; but not so new as they don't connect to our previous experiences.

Also, the concept of sexuality being leapfrogged - it's perhaps only the relatively modern concept of sexuality that needs to be leapfrogged. Scrolling back a few thousand years (a blip in human history), it is debatable whether the concept of sexuality even existed.

It looks like you have some interesting concepts. But I'm unable to follow any of it. If you try to make each sentence into a paragraph on its own, and then copyedit to get rid of the extra words, as well as to dumb down the phrases which (probably) sound sophisticated to you by two levels or more, I promise I will give it a lot of time and full attention. Pinky swear.

Is the android an effeminate space hero? Or someone else? What does the acclaim have to do with any of it? Stumbling on the very first sentence.

i am surprised nobody has taking the concept of shaders but used it for animated porn. i haven’t looked but you would assume it would be a thing by now considering how realistic video game graphics are becoming do to shading/light algorithms.

if you can textures basic animations of characters to have realistic skin, then that would be a major step towards ai porn, then the artist can draw however they want and a tool that would fill in the rest

There are some fairly realistic animated porn scenes. I think these are done by semi-professional artists (and probably involving a great deal of effort). But they tend to be... rather fantastical. So there's no mistaking that it is unreal.

Right now we can generate static images which will fool most people. And the wizards behind such magic could conceivably model physical bodies. In fact, this is done in movies quite often now, but not sexually (as far as I'm aware).

Well, I think I remember reading an article saying amatuer type stuff was some of the most popular. Ostensibly one could interpret this to mean that people want the realest stuff. I think a lot of this is cyclical, so maybe in 20 years or so we could see it swing back and have hyper real deep fakes all over the place.
But this is a problem of free markets. What was amateur in the past was free. What is amateur now is tied to some additional paid content.

I do not want to take away income from awesome people who choose to do that, but I also don't like fakes. Anyone who is producing content for money is or will eventually be scripting and producing the content. Thus, it will no longer be real, amateur. It might be low quality (and amateur in that regard), but it will not be real people enjoying real life with no other consideration at that moment.

So if everything is leading toward fake, perhaps we should just focus on building believable fake content to our ideal specifications.

I'm sure people are working on it, but it's still a cost issue.

Currently, a few min of animation at the level of a AAA video game is going to cost you many thousands, more than $100k if you're modeling off of human actors (which games do) and starting from scratch.

By contrast, you can pay a desperate person less than $10k to film hours of porn, and the end result doesn't have a risk of an "uncanny valley".

Lol. Forgive me for laughing at the wordplay of "uncanny valley".

Indeed though, this question I posed arose from the realization that Reddit amateur nude posts have changed over time from being just people sharing to people steering viewers to an OF page.

Then I reasoned that if AI could produce an endless stream of "content", what would this mean?

The technology does not exist.

No one can do this, yet.

Anime, CGI, current shonky deep fake porn exists. So this will definitely exist one day.

There is lots of legit money to be made first on whoever can invent the tech. Anything that can do what you suggest will be able to do a lot of amazing stuff.

>if we can generate incredibly detailed and believable faces

We can't. Not with 'deep fakes' HN articles about AI influencers are not real. They are models or groups of models that are enhanced to look a certain way.

>So the question is, if we can generate incredibly detailed and believable faces, why not entire bodies with faces? This must exist already.

This already exists for VR, check out virt-a-mate. https://old.reddit.com/r/VAMscenes/. obviously nsfw.

you can have very detailed created women doing what you want in VR

I may be out of touch here. Is this all (directed) algorithmically created? If so... it is believable enough to fool most people. And if so, within a few years it could probably fool all but the most critical viewers.
I believe it is all created and scripted out by the user creating it.
The problem is thus:

People would prefer actual & real adult content over fake simulated versions. There will always be a market for the real stuff, because computer generated stuff feels weird to look at and enjoy. It's too simulated and false to arouse you properly.

(Of course there will always be a market for people who enjoy simulated stuff too, but I don't see it replacing real content).

Maybe if sex work gets legitimized to the point of unionization. Then income from porn acting will go up, and with it the replacement of porn acting with deepfake.
Back in '02 I started a project to create automated digital doubles of anyone. I was a developer working in feature film VFX, and had been doing actor replacements as a specialty. Long story short, by '08 I had an MBA, a global patent on automated actor replacement in filmed media, a team with Academy Awards for VFX, and working feature film quality technology.

However, '08 was the financial meltdown, and at this time nobody had ever heard of automated digital doubles or automated actor replacements, so my efforts to raise financing were met with dismay, disbelief, and ridicule. Demonstrating by creating digital doubles of VCs on the spot, I get accused of fraud. So bizarre. Slowly I manage to get traction with angels, only to have one of them "realize what this could do with porn" and they'd fixate on porn. No amount of debate would convince them that such a company would be a lawsuit engine, be hated by half the human population, and yet have no stable revenue because porn is extremely unstable, as well as filled with flat out criminal personalities. I end up disbanding that angel group and try again. Two more times I manage to get an angel investor team formed, only to have one of them fixate on porn and then all of them follow. The debates I was in felt like arguing with Lucifer.

While traveling through this arena, I met others pursuing actual porn startups. There is investment money there, if one want to have partners like Flint Publications, or Penthouse. They have tech groups, believe it or not.

I was trying to create traditional video and print advertising, but personalized, with you and your family/friends in the ads delivered to you. It is still a viable concept, and I still have working feature film quality technology to create them. However, every time I dip my toes back into this area I encounter irrational pornographic business fantasies, and I just do not want to play that game.